![Sample our Bioscience journals, sign in here to start your access, Latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5925/TOC-Subject-Sample-2021-Bioscience.jpg"})
Abstract
This paper introduces a generalization of the simulated annealing algorithm for global optimization. Simulated annealing has been successfully applied to a number of combinatorial and continuous optimization problems. The original approach has been significantly improved by introducing adaptive annealing schedules and annealing several copies of the problem in parallel. In this paper we make a further step and propose a generalized simulated annealing algorithm called Demon Algorithm. This algorithm is constructed in analogy to the action of Maxwell's Demon and has been motivated by an information-theoretic analysis of simulated annealing. The algorithm is based on an ensemble of identical systems that are annealed in parallel. The ensemble evolves according to a sequence of target distributions with the aim to end up in a distribution that is concentrated on optimal solutions. The evolution of the ensemble is based on collective moves. The algorithm is implemented for the problem of graph bipartitioning and its performance is compared with conventional simulated annealing and a downhill search algorithm.